Re: [Simulavr-devel] [patch #7034] display frequency in fool-proof way

[Knut Schwichtenberg]

For me this leads to the main question: What do we want to achieve with the
simulator - do we want to run algorithmic tests in any time universe or in
realtime? If we want to test a SW in principle there is not need to have the
proper timing like in the LCD. This leads also to the question if there is a
necessity for a scope because it can only show a swingig pin, but if the
frequency is proper can not decided (or difficult).

Klaus Rudolph wrote:
>>>>> Using higher precision vars in the simulator only slows down the 
>>>>> simulation.
>>>> A good point for single-cpu simulations.
>>> A single cpu simulation needs no frequency value. :-)
I think if you are testing an algorithm that is not time dependant there is
definitvely no frequency required. If you run mutiple CPUs with different
frequencies- what simulavrxx can do - you need something like a relation between
each. And if you want to check a PWM,... (clock related HW-actions) you need a
frequency to make it understandable.

>> Alas, we're wrong.
>> Even a single cpu can use four frequency sources at once:
>> A crystal that determines the MIPS.
If you count the clockcycles you know the MIPS.

>> The watchdog.
>> A crystal for the real time clock.
>> An RC oscillator that times flash and EEPROM writes.
These are CPU clock independant frequencies

>>
> Oh, yes! A single "controller" need this, the cpu not. :-)
Hey, does a controller have/need a CPU? Really? :-)